Last updated: May 14th, 2023
At the end of the day, we all want to get around town safely and quickly. American cities generally tackle that need with more space for more vehicles. However, as added lane-miles fail to reduce congestion (but create a plethora of financial, social, and environmental issues), there’s a growing movement to move beyond a car-first approach.
Improvements to battery technology have made electric bicycles and scooters an increasingly viable alternative. They get around terrain and fitness limitations of walking and (unassisted) cycling, and occupy only a sliver of the space of a car.
That’s simple geometry, and all else being equal, we’d expect it to reduce traffic congestion. But all else seldom is equal, so that intuitive conclusion is surprisingly hard to prove in real life.
To clarify the matter, researchers at Georgia Tech’s School of Public Policy did a first-of-its-kind study to estimate the direct impact of e-bike and e-scooter use on urban congestion in Atlanta.
While their findings aren’t definitive—more on that later—the study strongly suggests that shared micromobility devices reduce congestion by replacing car trips.
After the city of Atlanta suddenly banned e-bikes and e-scooters at night, the researchers found “increases in travel time of 9–11% for daily commuting and 37% for large events.” In other words, congestion worsened when e-bikes and e-scooters were banned.
This article might contain affiliate links. As a member of programs including Amazon Associates, I earn from qualifying purchases.
How did the Atlanta micromobility study work?
At the risk of oversimplifying, researchers dividend the city into several zones based on e-scooter and e-bike access, no-ride zone enforcement, transit and facility locations, and commute origin/destination pairs.
They then compared Uber Movement car trip data for 45 days before and after the no-ride zone was implemented. There weren’t any other noteworthy changes in the environment, so changes in car trips were mostly the result of changes in micromobility access. (It’s not quite that simple, but we’ll get to the caveats later.)
Why exactly does micromobility reduce congestion?
It’s a simple matter of space. Moving from a car to a scooter shrinks one’s footprint down to almost nothing, so there’s more space available for the cars that remain. And, depending on local laws and infrastructure, the individuals who make the switch can more or less avoid traffic jams altogether.
However, substitution can be hard to predict, and seldom works out as cleanly as theory indicates. Ever since shared e-scooters popped up, there’s been a long-running debate over whether they reduce car use or simply replace non-car modes like walking, cycling, and transit.
If it’s the latter, then a no-ride zone will only lead to more non-car trips, which wouldn’t affect car travel time. But if shared micromobility does replace car trips, then a no-ride zone will increase car trips within the same fixed space, so car travel time must increase as well.
So, working backward: increased travel times absent any other major changes means e-scooter and e-bike trips really did replace some car trips after all.
How generalizable are the results?
I interpret this study as strong but incomplete evidence that micromobility improves congestion. To put it another way, it helps dispel the notion that e-scooters just take away from foot/bike/transit trips.
It’s convenient that this study happened in Atlanta, which like most American cities is not bike-friendly on the whole. There are several multi-use paths around the urban core, and a smattering number of protected (yet disconnected) bike lanes downtown, but bike infrastructure overwhelmingly comprises painted “bike gutters” and sharrows. In other words, nothing about Atlanta—besides perhaps the mild weather—is uniquely conducive to e-scooter or e-bike trips.
By the way, note that travel time during the ban increased the most during stadium events. That’s to be expected: congestion generally increases at those times, so there are simply more potential e-scooter trips.
That’s partly possible because Mercedes-Benz Stadium is connected by better bike infrastructure than most of the city, so e-scooters (which are encouraged to use bike lanes in Atlanta) are a safer and more convenient option than for other destinations.
As far as I can tell from older Street View photos, neither this infrastructure nor the general surroundings changed during the intervention. The legality of riding—not its safety or general—was the only observable variable.
Why e-bike & e-scooter congestion reduction is hard to measure
The gold standard of empirical research is the randomized controlled trial, wherein one single variable is changed for random subjects. But congestion, like every other facet of public life, reflects a web of factors that often influence each other. It’s virtually impossible to establish causality by manipulating just a single one.
In social sciences, the next best thing is usually a “natural experiment,” meaning a more or less random event or policy change. The assumption of randomness is critical and can be hard to establish.
The no-ride zone was sudden and arbitrary, taking effect the day after it was announced. Leading up to the ban, there’s no reason to think scooters were the subject of more public debate in Atlanta than in other scooter-share markets. What’s more, shared micromobility services can be geofenced and remotely shut off, so compliance was necessarily high. So, in this case, I believe randomness is a reasonable assumption.
The point is not to dissect their methodology, but to highlight why definitive and noncontroversial answers are so hard to come by.
On top of the authors’ own caveats in their paper, I’d highlight a few questions that could explain the congestion effects of the no-ride zone or at least limit our generalizations. I’m not saying any of the following are true—in fact, I highly doubt it—but it’s fair to ask:
- Did Atlantans’ preference for scooters have to reach a certain threshold before there were enough trips that city officials would even contemplate such a ban?
- Did the novelty of shared e-scooters at the time (2019) create a passing infatuation that incentivized trips that would not otherwise have happened at all?
- Would the policy have had the same effect during daytime hours, when car traffic is heavier, transit stops get more frequent service, and a somewhat different group of people are traveling in the first place?
Ultimately, these results are a huge step toward answering a critical question for transportation planners, and they unequivocally point to the need for more micromobility experiments—and the infrastructure to make them accessible to all.